Over the past year we have attempted to characterize more completely the cognitive disturbances in schizophrenia. In particular we have begun work on the mechanisms accounting for failures in memory in schizophrenia. Patients' difficulties do not appear to be due to qualitative abnormalities in susceptibility to interference, encoding, or so called false memory problems. We have begun to computationally model the episodic memory impairments in schizophrenia in order to determine if they are due to general noise or a single stage in mnemonic processing. We found that one possibile explanation of our results involves defective encoding of context information (a function asigned to the parahippocampal gyrus in our model. We have examined disorganized speech in schizophrenia using various semantic processing paradigms. In general patients have difficulties not with the size of their lexicon but rather how they access it automatically, as evidenced in priming paradigms. Thus, we have devised a battery of novel experimental tasks to assess whether schizophrenic patients show a dissociation between lexical integrity and semantic abnormality. We have completed work on a study which compares the integrity of the internal representation itself to the integrity of activation among representations using various types of number priming and quantity processing. This technique circumvents problems in judging the relatedness of words. We have also begun to employ a computationally rich technique called "latent semantic analysis" which judges the coherence of schizophrenic discourse using reliable computer controlled methods. We have found odd associations and diminished coherence over various discourse lengths. Importantly the technique is highly reliable; because it was highly correlated with clinical ratings from interviews we also think that it is valid. Finally, a genetic study of schizophrenia with an emphasis on intermediate phenotypes is ongoing. We are using a large battery of neurocognitive measures to characterize this "intermediate" phenotype. We base this on the rationale that patients do not inherit schizophrenia per se but a variety of susceptibilities to cognitive impairments and their attendant neurophysiologic abnormalities. We have already found that some cognitive measures yield high relative risks that are not redundant with diagnosis.Moreover, we have identified a gene (COMT) that has an impact on the N Back through dopamine signaling. We have also identified a gene (BDNF) that has significant impact on human hippocampal function, including episodic memory. We have examined a third gene, called G72, that demonstrates epistasis such that its effect on cognition was amplified in a group of schizophrenia patients (in contrast to controls and siblings). This is the first such report in the literature.